Environmental Hazards, 2015Vol. 14, No. 2, 156–186,

http://dx.doi.org/10.1080/17477891.2015.1028018

Mapping social vulnerability to flood hazard in Norfolk, England

K. Garbutta*, C. Ellulb and T. Fujiyamab

aCentre for Urban Sustainability and Resilience, Department of Civil, Environmental and Geomatic Engineering, University College London, Chadwick Building, Gower Street, London

WC1E 6BT, UK; bDepartment of Civil, Environmental and Geomatic Engineering, Univers ity College London, Chadwick Building, Gower Street, London WC1E 6BT, UK

*Corresponding author. Email: k.garbutt.12@ucl.ac.uk

Abstract

In this paper, we present a method to assess social vulnerability through the creation of an Open Source Vulnerability Index (OS-VI). The OS-VI provides context to environmental hazards and allows NGOs and local agencies to better tailor services and provide targeted preemptive

vulnerability reduction and resilience-building programmes. A deductive indicator- based approach is utilised to incorporate a wide range of vulnerability indicators known to influence

vulnerability. Unlike many vulnerability indices, the OS-VI incorporates flood risk as well as the loss of capabilities and the importance of key services (health facilities and food stores) through the measurement of accessibility when determining an area’s level of social vulnerability. The

index was developed using open-source mapping and analysis software and is composed completely of open-source data from national data sets. The OS- VI was designed at the national level, with data for all proxy indicators available across the entirety of England and Wales. For

this paper, a case study is presented concerned with one English county, Norfolk.

Highlights

We produce an open-source vulnerability index.

Accessibility to health care found to be severely affected by flooding.

High vulnerability areas found to be disproportionately impacted by flooding.

Urban extent of an area found to increase its level of vulnerability.

Flood affected areas more likely to be composed of elderly, sick and poor.

Keywords: vulnerability; floods; hazards; mapping; GIS; open source

1. Introduction

Flooding is a significant concern for much of the UK and is recognised as a primary threat

by most local councils, with coastal flooding in particular considered one of the highest

priority risks across nearly all counties (Cabinet Office, 2013). The UK has a lengthy

history of flooding, from the 1953 East Coast storm surge that killed 307 people to the

series of floods that affected much of the country throughout 2012, 2013 and 2014 and that

caused in excess of £1billion damage and led to the death of 26 people. Flooding, as with

other hazards, can negatively influence location-specific vulnerability as well as the

existing sociocultural and response and recovery mechanisms (Wamsler, 2014) by, for

example, changing hazard patterns, increasing the number of vulnerable people or

amplifying the loss of urban fabric and assets (Wamsler, 2014). It is widely accepted that

social, economic and political patterns exist within society that result in some groups of

people living with an amplified state of vulnerability (Enarson, 2007; Morrow, 1999).

Socially vulnerable populations are often restricted in their ability to respond – often due

to increased likelihood of health problems (Marmot, 2005) or constrained economic

support (Morrow, 1999) – and more often than not, lack access to critical resources during

disaster events (Halden, Jones, & Wixey, 2005; Morrow, 2008). This often augments the

way in which individuals and their wider communities are affected by environmental

hazards as well as how they respond and recover.

A greater understanding of the spatio-temporal variances in vulnerability and knowledge

of where those viewed as more vulnerable are concentrated within communities, as well as

the wider socio-economic circumstances of those communities, are key to understanding a

population’s level of resilience to environmental hazards (Cutter & Emrich, 2006) and

improving response service capability (Nelson, Lurie, Wasserman, Zakowski, &

Leuschner, 2007). In order to provide practical identification and assessment of social

vulnerability, the many influencing factors must be assigned measureable numeric

indicators (Atteslander et al., 2008).

As identified by the United Nations Expert Working Group on ‘Measuring Vulnerability’,

the only expert approach to identifying socially vulnerable populations is the production

of a vulnerability index: an aggregated or composite measurement of selected proxy

indicators of vulnerability, be it mortality, morbidity or social capital, that determines a

numerical value representing the social vulnerability of a given geographical unit

(Birkmann, 2005). However, as will be shown in the literature review, the majority of

vulnerability indices that have been produced are largely reliant upon generic,

geographically broad indicators, such as percentage of population below the national

poverty line or ethnic composition, which do not provide the precision required for the

fine-resolution geography under examination here or that is required by local councils or

NGOs who need to adequately understand social vulnerability within an area to plan

appropriately. Such indicators have been shown by others to be able highlight areas of

vulnerability, but we argue that an array of specific indicators spanning the many aspects

of modern life – from accessibility and education to the strength of local economies – are

required to gain a more substantive knowledge of modern social vulnerability.

In this paper, we examine social vulnerability through the creation of an Open Source

Vulnerability Index (OS-VI). A deductive indicator-based approach is utilised to

incorporate a range of vulnerability indicators previously shown to define and influence

vulnerability. In addition to vulnerability indicators, the OS-VI incorporates the risk of

flooding when determining an area’s level of social vulnerability. Unlike many hazard

mapping studies, it is not just proximity to the flood zone that is included within the OS-

VI; consideration is paid to the loss of capabilities and the importance of key services

(health care, food stores and employment centres) with the inclusion of measures of

accessibility factored in to the OS-VI. The index was developed using open-source

mapping and analysis software and is composed completely of open-source data from

national data sets. The OS-VI was designed at the national level with a resolution of

approximately 1500 people per Census tract, with data for all proxy indicators available

across the entirety of England and Wales. For this paper, a case study is presented

concerned with one English county, Norfolk.

We first review past vulnerability analyses, focusing on the act of measuring and assessing

vulnerability: indicator selection and index weighting as well as the inclusion of

accessibility within vulnerability analyses. Building upon the critical review of the

vulnerability literature, we present the choice of indicators used and the mapping processes

utilised to construct the index. The resultant OS-VI maps are presented and discussed as

well as a case study looking at accessibility to hospitals. We conclude with reflections on

the work undertaken and the potential refinement of the index and its future use.

2. Literature review

This literature review will examine the trends within past vulnerability assessments, with

a particular focus on the use of indicators to quantitatively measure vulnerability and the

weighting of said indicators within vulnerability indices, followed by an examination of

the role of accessibility within social vulnerability and the methods used to measure it.

Although many more challenges and debates still exist that centre on the history,

conceptualisation and definition of vulnerability, we will not focus on them here and

instead point the reader towards more detailed examinations of the theoretical

developments of vulnerability within the academic literature (see: Adger, 2006; Blaikie,

Cannon, Davies, & Wisner, 1994; Cutter, 1996; Cutter, Emrich, Webb, & Morath, 2009;

Liverman, 1989).

2.1. Vulnerability assessments

Countries, counties and cities are not homogenous, but are instead made of unique

communities that represent the most affluent and privileged, as well as the most destitute

and deprived. Vari- ations in socio-economic and health vulnerabilities exist in space and

time and indicators are available that are regularly and systematically collected at a

resolution small enough to allow quantitative measurement and mapping of such variations

at a community level (Cutter, Burton, & Emrich, 2010).

A vulnerability perspective has been shown to focus the attention of social research on the

diversity within populations; concentrating on the broader social, cultural and economic

factors that influence a community’s vulnerability and moving focus away from a singular

examination of the presence of hazards (Van Zandt et al., 2012). Vulnerability assessments

are often the first step to identifying those groups in society who may require added

assistance during an emergency (Blaikie et al., 1994; Morrow, 1999; Müller, Reiter, &

Weiland, 2011). However, due to its somewhat ambiguous nature, many assessments of

vulnerability have, to a certain degree, been vague, oversimplified or ill-structured

(Kværner, Swensen, & Erikstad, 2006; Müller et al., 2011).

Most vulnerability assessments examined were found to be concerned with locating

physical structures or population concentrations at risk within known hazard-prone areas

(Kværner et al., 2006; Müller et al., 2011). However, as events such as Hurricane Katrina

showed, the nature of the population – race, ethnicity, wealth, employment, income, and

so on – is just as important a factor as where that population resides (Long, 2007; Oxfam

America, 2009; Tate, 2013).

The majority of the research into vulnerability assessments undertaken over the past 20

years has focused on developing frameworks that are applicable to various systems, scales

and hazards (Stânga & Grozavu, 2012; Taubenböck et al., 2008). However, scale and

versatility have been identified as major constraints of many vulnerability assessments

(Preston, 2012). Much of the literature examined was found to focus on, and assign great

premium to, economic aspects of life, namely income (Preston, 2012), and to limit research

by geography (state, territory or census block) and the range and applicability of indicators

used. Relatively few studies have been operationalised in real-world settings (Cutter et al.,

2009), with fewer still replicable outside of their original focal zone. Assessments are often

designed with a focus on a specific geo- graphical area, often limiting their universal usage

(Engle, 2011), or the methods and data used often render the indices inappropriate or

incompatible with other areas, timescales or systems (Lankao & Qin, 2011). Further,

limited attention was found to be paid to open-source data usage, with proprietary data and

technology utilised widely. Few studies were found to have produced an index that could

be freely and easily utilised and adapted by NGOs and communities (Garbutt, 2013). Due

to these problems of comparability between indices, variables and methods, vulnerability

assessments are best thought of as an heuristic illustration of the conditions within the study

area, be it existing or anticipated, and not as a predictive tool (Cutter e t al., 2009). In terms

of scale, vulnerability assessments are generally undertaken at one of three levels:

household, regional or national. Many studies exist that examine relative vulnerability

across countries (Webber & Rossouw, 2010), such as Turvey’s (2007) composite

vulnerability index (CVI) of 100 developing countries, but rarely is attention paid to the

differing vulnerability of regions within these countries. At the local level, there is a large

and growing body of work examining local-level vulnerability, be it at the household level

(Bird & Prowse, 2008) or Cutter’s place-based vulnerability indices centred upon United

States Census Bureau block-groups (Cutter, Mitchell, & Scott, 2000) or later community

resilience studies undertaken at the larger county level throughout coastal US states (Cutter,

2006; Cutter, Barnes, & Berry, 2008).There is a growing body of literature on

vulnerability science that seeks to increase the usefulness and use of vulnerability

assessments, with an academic focus on quantitative measures of vulnerability and the

developments of a universal metric or measurement tool for vulnerability assessments (see:

Cutter et al., 2009; Jones & Thornton, 2003; Kasperson & Kasperson, 2001; Polsky, Neff,

& Yarnal, 2007). However, much of the literature reviewed also advocates a fine-

resolution, place-based, assessment methodology in an attempt to produce more local and

relevant observations (Barnett, Lambert, & Fry, 2008). 2.1.1. Indicators of vulnerability

Indicators – ‘quantitative measures intended to represent a characteristic or a parameter of

a system of interest’ (Cutter et al., 2008, p. 7) – are regularly used within vulnerability

science to produce a single universal value that represents vulnerability at a particular

temporal or spatial scale or location. A wide variety of indicators and assessment models

are in use with common indicators including: mortality, morbidity, social capital and

physical assets (see: Environmental Sustainability Index (Esty, Levy, Srebotnjak, & de

Sherbinin, 2005), Human Development Index (Anand & Sen, 1994; Burd-Sharps, Lewis,

& Martins, 2008), Human Well-being Index (Prescott-Allen, 2001) and Social

Vulnerability Index of Climate Change in Africa (Vincent, 2004)). The proxies available

for each indicator are yet more numerous (Rygel, O’Sullivan, & Yarnal, 2006).

A wide-ranging selection of relevant vulnerability indicators is presented within the

literature. Cutter, Boruff, and Shirley (2003) alone present a version of their index that

utilises 42 indicators. However, this project was concerned with vulnerability in a

rural/urban region within a developed country, which means that not all vulnerability

indicators identified within the literature are appropriate. For example, common indicators

used within vulnerability indices are measures of political stability, access to education and

literacy rate (Mustafa, Ahmed, Saroch, & Bell, 2011; Patt, Tadross, Nussbaumer, Asante,

& Metzger, 2010). These indicators are arguably inappropriate for use within developed

countries. Other indicators utilised in earlier studies, such as social-capital religion, local

asset value and civic involvement (Müller et al., 2011), are unsuitable due to limited data

availability.

Challenges to the production of vulnerability metrics, especially a universal index, and the

use of such indicators do exist, with vulnerability indices limited by the lack of consensus

on a strict definition of vulnerability across the discipline; selection of indicators;

determination of indicator importance; data availability, quality and validation; and

difficulty in quantifying social inter- actions and measuring concepts such as institutional

capacity or readiness (Müller et al., 2011). Furthermore, indicators, like models and

indices, represent reality imperfectly. The reliance upon measurable data, both quantitative

and qualitative, to produce a vulnerability index limits it to those vulnerability components

that are quantifiable (Moldan & Dahl, 2007). Interactions and feedbacks exist that are not

sufficiently understood and thus cannot be factored into the production of accurate

indicators (Damm, 2010). The impact of these ‘non-quantifiables’ must be gleaned from

qualitative research, working with NGOs, for example, to further elucidate the find- ings

from studies like the one presented herein.

To be effective, vulnerability indicators must be appropriate, representative of the

phenomenon under examination, credible and feasible (Moldan & Dahl, 2007, Niemeijer,

2002). However, methodological trade-offs are often necessary and a balance between

these dimensions, as well as scale and cost, is required to produce legitimate and usable

indicators (Damm, 2010). As with using too broad a geographic scale of analysis – county

over Census tract, for example – the use of general, broad-based indicators can reduce the

effectiveness of the index as populations under analysis are grouped and classified too

coarsely, perhaps even stereotyped (Goss, 1995), and the nuances of vulnerability and local

minutiae are lost. Similarly, vulnerability represents social geography (Cutter, 2006) and,

as such, indicators must vary with changing geography (Rygel et al., 2006). Indicators infer

knowledge of a system (Balica & Wright, 2010) and to fully understand the vulnerability

system, we believe, the indicators used should be as locally relevant as possible.

2.1.2. Weighting

The process of weighting vulnerability variables is often idiosyncratic and is the subject of

much debate (Bohringer & Jochem, 2007; Brooks, Adger, & Kelly, 2005; Morse, 2004;

Yoon, 2012). Two camps exist whereby studies either treat all factors equally within an

index, or they apply weightings based upon perceptions of indicator importance, often with

weightings assigned with little to no input from those within the society being examined

(Dibben et al., 2004).

Many vulnerability assessments develop weighting schemes and believe that this is

necessary to accurately assess vulnerability as not all elements within society play an equal

role in creating, fostering or reducing vulnerability (Haki & Akyurek, 2004; Meyer, Haase,

& Scheuer, 2007). For example, the English Indices of Deprivation, which provides a

relative measure of deprivation across England at the Lower Layer Super Output Area

(LSOA) resolution, assigns a combined 45% weighting to issues of income and

employment and 13.5% to issues around health and dis- ability. Similarly, in developing

their weighting schemes, Rygel et al. (2006) ranked the factors within each indicator to

assign importance; Cox, Rosenzweig, Solecki, Goldberg, and Kinney (2007) assigned

weightings based on the per cent variance explained by each factor; Brooks et al. (2005)

employed focus groups to identify vulnerability indicators, as well as relying on those

previously identified by earlier studies, and to provide weightings for each indicator

chosen.

However, many vulnerability studies examined preferred not to assign weights to their

chosen indicators, claiming that they are of equal importance to the calculation of

vulnerability or are independent of each other (Yoon, 2012) or that insufficient evidence

exists to accurately assign importance to any one demographic factor over another to

produce a robust weighting system (Fekete, 2011). However, Fekete (2011, p. 1167) does

acknowledge that ‘the avoidance of weight- ing is in itself a certain kind of weighting ...

because not weighting means equal weighting within the aggregation of input variables or

indicators’.

2.2. Accessibility

Accessibility to key services is severely affected during an emergency, particularly during

flood events where people can be stranded in their homes, roads can be washed out, and

services damaged or stopped, further compounding the vulnerability of individuals or

groups, particularly the elderly. Although the immediate health impacts of floods (deaths

and injuries caused by drowning) are relatively rare in the UK compared to the medium-

to long-term effects (particularly psychological stress) (Reacher et al., 2004; Tapsell &

Tunstall, 2006), timely access to a hospital during such an emergency situation remains a

requirement that can reduce a person’s overall level of risk (Campbell et al., 2000; Nicholl,

West, Goodacre, & Turner, 2007). However, according to the Department for Transport

(DfT), the UK Government department responsible for the majority of the UK’s transport

network, accessibility to health care appears to be getting worse, with less than one-third

of the population having ‘reasonable’ access to a hospital (defined as travel time under 30

minutes) (DfT, 2012). The most striking element of the DfT accessibility data are the

differences between the UK’s rural and urban communities: rural resi- dents are required

to travel for up to 29 minutes on average to access key services (hospitals, food stores,

employment centres, schools and town centres), compared to an average of 12 minutes for

urban residents. This is an issue of growing concern, given the increasing rural popu- lation,

approximately 10 million, whose residents have a far higher age profile than their urban

counterparts as well as an increasing number of residents with an activity limiting health

problem or disability (Office for National Statistics, 2013). It is believed that more than

one million UK residents annually do not seek medical assistance and/or miss medical

appointments due to limited accessibility, severely impacting an individual’s health and

compounding social exclusion (Nettleton, Pass, Walters, & White, 2007).

Although a significant body of literature exists examining accessibility from a number of

per- spectives (see: Higgs, 2004; Liu & Zhu, 2004; Mavoa, Witten, McCreanor, &

O’Sullivan, 2012), much of the academic work on accessibility has focused on the impact

of physical restraints or barriers, for example, topological or spatial barriers (rivers or

mountains) or the impact of car ownership (or lack thereof), whereas the examination of

temporal constraints on access to key services (the time taken to reach a hospital) has been

the subject of far fewer empirical studies (Delafontaine, Neutens, & Weghe, 2012). Recent

theoretical advances and improvements in the quality and availability of data have led to

an increased interest in the application of accessibility as a measure of vulnerability (Páez,

Scott, & Morency, 2012). Areas such as social exclusion (Ribeiro, Antunes, & Páez, 2010),

regional and urban planning (Comber, Brunsdon, & Green, 2008), the economy of

transportation infrastructure (Páez et al., 2012) and the location of healthcare facilities

(Coffee et al., 2012; Ngamini Ngui & Vanasse, 2012) have all seen a growing interest in

accessibility. The causal relationship between transportation disadvantages and social

exclusion has become central to most modern transport planning (Chen, Keith, Air riess,

Li, & Leong, 2007; Neutens, Delafontaine, Scott, & De Maeyer, 2012; Preston & Rajé,

2007; Scott & Horner, 2008). Despite its importance, few vulnerability assessments were

found to have included a measure of accessibility within a vulnerability assessment.

2.3. Literature review conclusion

Findings from this literature review have demonstrated the breadth of vulnerability analysis

and the continued desire to quantify vulnerability and the many methodologies employed.

However, the review revealed various widespread methodological limitations. Past work

on measuring and mapping vulnerability was found to suffer from four main problems: (1)

the focus on population concentrations within known hazard zones and not entire

populations, potentially omitting vulnerable populations; (2) it has been limited by scale –

either too large to uncover local-level nuances or too insular to allow for useful

comparisons at a country or larger level, often utilising site-specific data or indicators; (3)

the focus on income as the key variable for measuring vulnerability, deprivation or

resilience; and (4) the reliance on proprietary data and/or methodologies.

The review found a plethora of large-scale (country to near-global) attempts to assess

vulnerability with a focus on relative measurements of vulnerability based predominantly

on the economic aspects of life and idiosyncratic views on indicator weighting. Few studies

that took account of social aspects of vulnerability (health, well-being and support) were

found to realistically examine vulnerability on a local scale using an indicator-based

approach.

The review identified few vulnerability assessments that included a measure of

accessibility within the measurement of vulnerability, despite the recognition of its

importance. Limited quantitative research was found that combined geo-demographic

analysis of vulnerability and hazard mapping to produce a CVI at a national scale with a

resolution smaller than county or region level. A lack of integration between academic

analyses of vulnerability and governmental and non-governmental policy was also noted

(Mustafa et al., 2011), with the requirements, scale, complexity and enquiry method

required by both such parties being often at odds (Mustafa et al., 2011).

3. Methodology

3.1. Study area

Norfolk is a low-lying English county with an extensive coastline – 93 miles. Norfolk has

a lengthy history of coastal and riverine flooding, ranging from the North Sea flood of 1953

to several less extensive, but still damaging, flood events occurring over the past half

century. It is estimated that 100,077 properties in the county are at risk from flooding, with

the areas of Norwich, King’s Lynn and Great Yarmouth highlighted at particular risk

(Carroll, 2012). It is an area with a near-equal rural/urban divide with an aged population

and a low population density – 160 people per square kilometre (compared to the UK

average of 407). Norfolk has limited accessibility options despite being one of the largest

counties in the UK: it has no motor- way, direct access to only three primary ‘A’ roads

(A11, A12 and A47), only one major railway station and has only two NHS hospitals with

accident and emergency facilities (Norfolk & Norwich University Hospital and Queen

Elizabeth Hospital) (Figure 1).

3.2. Geography and map dataChoosing the correct level of Census geography is essential

to highlighting the real patterns and issues within large governmental data sets. It can be

detrimental to simply choose the smallest Census tract available. When looking at fine-

resolution geographies, particularly Output Area (OA) and postcode level, confidentiality

becomes a problem as data supplied are likely subjected to rounding to ensure anonymity

when dealing with, for example, instances of violent or sexual crime or users of mental

health services. In addition, choosing a level of geography such as post- codes results in a

vast increase in the level of investigation and calculation required, adding strain to the GIS

produced (for example, there are 2.5 million postcodes in the UK compared to 171,372

OAs). Conversely, choosing a medium–large geography, Medium Super Output Area or

Local Authority (LA), would speed up investigation and allow for easier integration of

health and crime data, but reduce the applicability of results at a community/neighbourhood

level, as a single vulnerability score would be produced for in excess of 15,000 people.

Examination of the different geography products available led to the decision to use the

LSOA boundaries for final analysis and visualisation (Figure 2). LSOA has a minimum

population of 1000, with an overall mean of 1500, providing a sufficiently fine-grain detail

to examine

Figure 1. Study area map showing the location of the county of Norfolk and its major urban

towns.

Figure 2. A comparison of ONS census geography/boundaries.

vulnerability at a community level. In addition, key data sources, namely Office for

National Statistics (ONS) Census, readily supply data at the LSOA level and the resultant

LSOA-based OS-VI would be easily compared to the English Indices of Deprivation

(Index of Multiple Deprivation (IMD)) and Experian’s MOSAIC classification (see

Section 5.2). There are 539 LSOA in Norfolk, 34,753 throughout England and Wales.

This project takes full advantage of the open-source community and utilises a wide range

of open-source digital maps supplied by the Ordnance Survey (OS) and ONS and key

service location data provided by OpenStreetMap (OSM). It was hoped that OSM road data

could be utilised for routing purposes. However, when OSM road data were compared to

OS OpenData data, a significant number of roads, particularly local residential streets and

small country tracks, were either missing or significantly misaligned and lacking meta-data

required for robust routing. Thus, it was decided to use an OS MasterMap Integrated

Transport Network (ITN) road data set, freely available from the OS under their

OpenDataTM initiative to develop a routable road map.

3.3. Flood data

UK flood data, specifically information pertaining to flood affected areas and flood

defences, is freely available via the Environment Agency’s website. It was decided that,

for this project, the type of flooding (tidal, fluvial or ground) was not required for analysis,

just the knowledge that a certain area was or was not within a flood zone. It was decided

to plan for the ‘worst- case scenario’ and utilise the ‘extreme flooding’ scenario: 0.1% (1

in 1000) or greater chance of happening each year (Figure 3). In addition, the LSOA

Centroids, which represent a summary single reference point of how the population at

census time was spatially distributed and grouped within that LSOA, were utilised to define

those centroids within the flood zone. This was utilised within accessibility analysis and as

part of a hazard indicator.

3.4. Accessibility mapping and key services

As stated in the Literature Review, access to key services is an important aspect of

vulnerability and the loss of capabilities that accompany limited access to key services is

well known (Miller, 2003). The UK Government states that 30 minutes is a ‘reasonable’

time to access a key service (DfT, 2004). Thus, to gain a greater understanding of

accessibility within Norfolk, key services within the study area, including hospitals,

general practitioner (GP) surgeries, large food stores and schools, were mapped using data

from OSM.

Travel time to a hospital was chosen for analysis not due to the potential increased demand

for hospital access caused by the flood event – there is little evidence in the UK to suggest

that flooding leads to an increase in hospital demand in the immediate aftermath of a flood

(Bennet, 1970; Floyd & Tunstall, 2004; Tunstall, Tapsell, Green, Floyd, & George, 2006)

– but due to the vital health, well-being and social care services that hospitals provide to

the rural and largely elderly communities within the study area. Any reduction or disruption

to these services due to reduced service provision as emergency response work takes

priority (Kazmierczak & Cavan, 2011) and routine social care personnel respond to those

impacted by the floods (World Health Organization [WHO], 2002) or the loss of service

due to hospital closure or damage (Greater London Authority, 2013; WHO, 2002) or

reduced accessibility due to restricted transport options or damaged trans- port systems

(Aday, 2001; Morath, 2010; WHO, 2002) could exacerbate pre-existing community

vulnerabilities, health concerns and the stresses of flooding (Hajat et al., 2005).

Service Area and Closest Facility analysis was undertaken using QuantumGIS and the

routing metadata within an OS ITN road data set (Figures 4 and 5). It was assumed that all

journeys started at the LSOA Centroid and were taken in a car travelling at the maximum

speed allowed. The impedance was time and the fastest routes were calculated. Analysis

was repeated under flooded conditions where it was assumed that, under the ‘worst-case

scenario’, all roads within the flood zone would be impassable and that key services would

likely be damaged or inac- cessible. Thus, roads and key services within the flood zone

were restricted. Accessibility results fed into the OS-VI, with any LSOA registering travel

time to a key service above the average receiving a score of one and all others a score of

zero. A case-study example of those LSOA that were unable to reach a hospital due to the

flood zone restrictions is presented.

Figure 3. Norfolk local authorities with flood zone overlay.

Figure 4. (a) Service area – 30 minutes travel time from hospitals under non-flooded conditions

and (b) Service area – 30 minutes travel time from hospitals under flooded conditions.

Figure 5. (a) Closest facility analysis to nearest hospital under non-flooded conditions and (b)

closest facility analysis to nearest hospital under flooded conditions.

3.5. Identification and analysis of vulnerability indicators

The selection of indicators had two controlling factors: (1) justification of its relevance to

the measure of vulnerability based upon extant data/literature and (2) the availability of

open- source data presented at an appropriate geography that is of consistent quality.

Although there is no agreement on a single set of vulnerability indicators within the wider

literature, there is agreement that vulnerability stems from elements related to economic

and material wealth, health, institutional support, accessibility and the presence of hazards

(Bruneau et al., 2003; Gunderson, 2009). Subsequently, through literature analysis and

focus group discussions with experts from NGOs and community groups affected by

flooding, a ‘short list’ of 77 key vulnerability indicators was produced that met the

controlling factors. Further discussions reduced the list to a set of four category elements,

which are further divided into sub-categories and finally proxy variables and their

respective indicators. In total, 53 different vulnerability indicators were chosen to form the

OS-VI (Table 1). Data were acquired, examined and prepared as a full England and Wales

data set and filtered for the Norfolk case study herein.

3.6. Development of a vulnerability index

A vulnerability ‘score’ was produced for each of the 539 LSOAs of Norfolk, England. All

indicators were normalised, translated into scale-free relative frequencies per LSOA and

reduced to a binary format: either a zero or a one, with a zero representing no vulnerability

and a one representing the presence of vulnerability. The indicator for each vulnerability

variable is based upon the average figure for that variable within England and Wales as a

whole, that is, for the variable ‘percentage of population of LSOA aged 16 or under’. If the

result is above the national average, the LSOA is assigned a binary vulnerability score of

one for that variable, or if the result is below the average, it is assigned a zero. It was felt

that the use of an average for each vulnerability indicator (both national and county) offered

a suitable measure of potential vulnerability by representing a baseline whereby those

below the average are arguably more vulnerable than those above the baseline.

It was decided that a transparent and easy-to-understand indicator system would be best,

given the intent to use the system outside of academia. Whilst a plethora of weighting

methods exist that are subjective or reliant upon data analysis, such methods do not

adequately reflect the priorities of decision-makers (Cutter et al., 2010; Esty et al., 2005).

As such, an equal-weighting system was utilised for the OS-VI, but team leaders from the

consulting NGO were provided with an interactive ‘dashboard’ whereby they could define

the weightings used for the calculation of each vulnerability category so that they could

produce an index that represents the priorities of their department.

Two separate OS-VI were produced, one based upon variable averages for England and

Wales as a whole and another that examines Norfolk in isolation. This was done to provide

information on vulnerability in Norfolk in relation to that of England and Wales but also

to identify vulnerability within Norfolk under a local context. The intention of the National

OS-VI was to present the vulnerability of Norfolk in relation to England and Wales, not in

context to England and Wales. What the presented National OS-VI does not show is the

rating of any given Norfolk LSOA in relation to the other 34,214 LSOA in England and

Wales. Only Norfolk LSOA was ranked. It was out of the scope of this project to produce

an entire National OS-VI. Although several variables were processed at the national level,

several key variables, namely those relating to flooding and accessibility, could not be

processed to the extent necessary within the given timeframe. The Local OS-VI was

created using average figures for Norfolk only. This localised OS-VI provided a more

place-based examination of vulnerability.

3.7. Visualising the OS-VI

OS-VI results were ranked (LSOA with the lowest cumulative vulnerability score ranked

1 and the highest score ranked 539), divided into four vulnerability ratings and, following

discussions with NGO sponsors and beneficiaries, visualised in the near-universal and

easily understood traffic light style graduated symbology that allows for quick

interpretation and is widely recog- nised within risk and emergency management sectors

(see Cabinet Office, 2013):

1) low vulnerability (1–134);

2) low to moderate vulnerability (135–269);

3) moderate to high vulnerability (270–404) and

4) high vulnerability (405–539).

As is given, areas rated high indicate areas with a relative higher level of vulnerability

because a high number of assessed variables were above average for that LSOA in the

National or Local OS-VI, suggesting that that LSOA is in a more vulnerable state relative

to the average National or Norfolk LSOA. Similarly, areas rated low indicate areas with a

relative lower level of vulnerability and suggest that LSOA is in a less vulnerable state

relative to the average National or Norfolk LSOA.

4. Results

4.1. The open-source vulnerability index

As can be seen in Figures 6 and 7, the National and Local OS-VI display a similar

vulnerability distribution. No significant trend was noted that could explain the changes

between the two OS- VI. Both OS-VI present vulnerability within Norfolk as following a

general radial pattern around the major urban areas, namely the city of Norwich, with

vulnerability high within the urban centre and decreasing outward. A low vulnerability

‘ring’ can be seen to encircle Norwich, representing the relatively affluent suburbs of

Norwich.

The OS-VI scores range from 12 (low vulnerability) to 45 out of 53 (high vulnerability)

for the full National OS-VI, with a mean score of 24, and a range of 6–45 out of 53 for the

Local OS-VI, with a mean of 22. With the exception of two LSOA, the most vulnerable

LA is Great Yarmouth, with approximately 85% of LSOA in both indices rated high or

moderate-high. Topping off the list as the most vulnerable LSOA is Southtown & Cobholm

in Great Yarmouth (E01026635), with a vulnerability score of 45 out of 53 in both indices.

This is expected given the area’s deprived economy and housing, as well as the general

poor health of residents and the presence of a flood hazard across 25% of its area. In

comparison, the least vulnerable LA is Broadland, with 91% and 86% of LSOA in that area

rated low or low-moderate in the National OS-VI and Local OS- VI, respectively. The least

vulnerable LSOA within the National OS-VI was Town (E01026945) in South Norfolk

with a score of 12 out of 53. Sprowston Central (E01026556) in Broadland is the least

vulnerable LSOA in the Local OS-VI, with a vulnerability score of 6 out of 53.

Table 2 shows the variables that contribute the most to the vulnerability scores recorded

within the England and Wales OS-VI. As can be seen, all 539 LSOA within Norfolk

recorded a drive time to a food store in excess of the national average. This was not

unexpected, given the study area’s rural geography, but does represent a major source of

vulnerability, particularly given the elderly nature of the area’s population (390 LSOA

recording above-average elderly populations) and its poor health and mobility (366 and

326 LSOA recording above-average number of residents report- ing limited actions and

long-term health problems/disability, respectively).

The LSOA with a high vulnerability rating are characterised by a disproportionate mix of

urbanised towns or cities, 70%, compared to rural towns or villages, which account for just

30%. In comparison, the urban–rural divide within those LSOA with a low vulnerability

rating is roughly equal, 49–51%. For those areas with a low-moderate or moderate-high

rating, the urban–rural divide is approximately 40%/60%. Norwich, which is entirely

urbanised, recorded only one LSOA with a low vulnerability rating within the National

OS-VI, with 73% of Norwich LSOA recording moderate-high or high ratings.

4.2. The impact of flooding on vulnerability

In total, 161 of 539 LSOA (30%) in Norfolk were found to be impacted by flooding and

saw their OS-VI ratings increase: 110 LSOA were cut off from hospitals during the flood

scenario; 57 had more than 50% of their area within the flood zone; and 33 LSOA Centroids

were within the flood zone. Only six LSOA recorded all three flood risks, all within the

LA of Great Yarmouth and Kings Lynn and West Norfolk.

Excluding those within the urban centre of Norwich, the major clusters of LSOA with a

high vulnerability rating can be seen to loosely match those areas with greater exposure to

the flood zone (Figure 8). Great Yarmouth in the East and Kings Lynn and West Norfolk

to the West account for 75% of those areas that received a high vulnerabili ty rating and are

impacted by flooding. Analysis found that 40% and 37% of those LSOA with a high or

moderate-high vulnerability rating were impacted by flooding in some way, compared to

just 20% and 22% of those with a low or low-moderate rating in the National and Local

OS-VI, respectively. Furthermore, 79% and 90% of those LSOA with the majority of their

area within the flood zone recorded a high or moderate-high vulnerability rating in the

National and Local OS-VI, respectively.

4.3. The impact of accessibility on vulnerability

Accessibility analysis found that travel time to a hospital in normal, non-flooded,

conditions ranged from 1 to 62 minutes. Norwich was the most accessible area in terms of

drive time, with most key services reachable within an average of 12 minutes. Breckland

and Great Yarmouth were the least accessible LA: travel time ranged from 19 to 62

minutes, with an average of 41 minutes in Breckland; and 90% of LSOA recorded travel

time in excess of 30 minutes in Great Yarmouth. The LSOA Suffield Park (E01026778) in

North Norfolk recorded the lowest drive time under both ‘normal’ and ‘flooded’ scenarios,

just 1 minute.

Figure 6. (a) Local OS-VI and (b) National OS-VI.

Figure 7. (a) Proportion of each LA in local OS-VI by OS-VI rating and (b) proportion of each

LA in national OS-VI by OS-VI rating.

Table 2. Variables contributing most to the vulnerability score.

Figure 8. National OS-VI with flood zone overlay.

Under the flood scenario, when all roads within the flood zone were restricted, 110 of 539

LSOA could not reach a hospital, 38% of these were in Kings Lynn and West Norfolk. One

hundred and fifty LSOA recorded travel times in excess of 30 minutes during normal

conditions. This increases to 179 during flooded conditions, or 289 when LSOA with no

access are included. Two hundred and thirty-six LSOA recorded no change in travel time.

The LSOA with the largest change in travel time between ‘normal’ and ‘flooded’

conditions, and subsequently the longest overall travel time, was Watlington in Kings Lynn

and West Norfolk (E01026722), which saw a 104-minute increase, from 22 to 126 minutes.

Accessibility in Norfolk is largely controlled by the rural/urban make-up of each area. Fifty

per cent of Norfolk LSOA are listed as rural, with 59% of those listed as being dispersed

or in a sparse setting. Travel time to a hospital in rural LSOA increased by 62% from an

average of 29 minutes during non-flooded conditions to 47 minutes during the flooded

scenario. For urban LSOA, travel time increased by 29% from an average of 17 to 22

minutes.

Analysis found that average travel time to a hospital across Norfolk as a whole under

flooded conditions increased by 52%, from an average of 23 to 35 minutes. Norwich was

again the most ‘accessible’ area, with average travel time remaining at 12 minutes and all

LSOA available to reach a hospital. In comparison, Kings Lynn and West Norfolk saw the

largest travel time increase: recording a 223% increase from an average of 21 to 68 minutes.

In addition, 48% of Kings Lynn and West Norfolk LSOA Centroids could not reach a

hospital. Great Yarmouth remains the least accessible LA, with 57% of Great Yarmouth

LSOA unable to reach a hospital during flooded conditions and the local hospital, the James

Paget University Hospital, largely cut off by the flood zone (Figures 4 and 5). The number

of LSOA in Great Yarmouth without reason- able accessibility increases to 90% when

those with travel time over 30 minutes are included.

5. Discussion

5.1. The demographics of flood vulnerability in Norfolk

The OS-VI not only displays the flood zones and therefore those areas at increased risk of

flooding, it provides a means of highlighting those populations that are potentially more

vulnerable to the impact of flooding due to their demographic characteristics. We term this

the demographics of flood vulnerability: populations that are potentially more vulnerable

to the identified flood risk due to its potential to exacerbate the local vulnerabilities

identified by their demographics. This may be due to the area having a high proportion of

elderly people or residents with health problems or the potential loss of key services during

flooding, or, more likely, a combi- nation of such vulnerabilities.

Flooding was found to impact one-third of LSOA within Norfolk and a link between the

presence of flood hazard in an area and its overall vulnerability was noted. Those areas

with a high or moderate-high vulnerability rating are disproportionately affected by

flooding, whereas only four LSOA with a low vulnerability rating were found to have the

majority of their area within the flood zone. Furthermore, of those LSOA impacted by

flooding, our analysis suggests that residents are also more likely to live alone and be aged

65+; be retired; have an income below the national median; be in receipt of a key benefit;

lack central heating in their home; have bad or very bad health; have limited actions due to

a long-term health problem/disability; provide care to another in excess of 50 hours a week;

and are more likely to live in an area where travel time to key services is in excess of the

national average. This suggests an underlying causal relationship between proximity to the

hazard and socio-economic and health vulnerabilities – a trend noted by other authors (see:

Alexander, 1993; Blaikie et al., 1994; Watts & Bohle, 1993) – although the relationship is

unclear and further study is needed.

A rural/urban divide within vulnerability was noted also. Accessibility has already been

shown to be impacted by an area’s rural or urban characteristics , with flooding further

exacerbating rural area’s limited accessibility. However, the link between an area’s

rural/urban make- up and its vulnerability is not as clear. A predominantly rural area’s

vulnerability appears to be more changeable. For example, North Norfolk, which is 85%

rural, recorded approximately 40% low or low-moderate ratings; whereas South Norfolk,

which is 68% rural, recorded approximately 80% low or low-moderate ratings. However,

predominantly urban areas were found to be dominated by high vulnerability ratings. For

example, Great Yarmouth, which is 64% urban, recorded approximately 80% high or

moderate-high ratings and Norwich, which is completely urbanised, recorded

approximately 75% high or moderate-high ratings. This implies that high vulnerability in

an area is to some degree related to its urban extent: increased urban space leads to higher

levels of vulnerability. However, not all urban areas are equally vulnerable, nor are the

underlying vulnerabilities the same. More work is needed to uncover the relationship

between rural/urban extent and the factors that influence vulnerability.

The LA of Norwich, which represents the only major city in Norfolk and the most populous

region, demonstrates the rural/urban divide of vulnerability well. Within Norwich, the

majority of LSOA are rated either moderate-high or high within both OS-VI. This high

vulnerability urban pattern (Figure 9) is similar amongst both OS-VI and is a phenomenon

noted within the literature whereby urban centres throughout the world are found to have

high rates of vulnerability, as well as deprivation, and are often surrounded by more

affluent suburbs with considerably lower levels of vulnerability (Erskine, 2010; Gartner,

2011; Musterd & Ostendorf, 2013). This is evidenced by the two LA that encircle Norwich,

Broadland and South Norfolk, both recording much lower levels of vulnerability within

both OS-VI (Figures 3 and 6).

5.2. Index comparison

When both the National and Local OS-VI are compared, there is little variation within the

overall distribution of the vulnerability scores. No significant trend in the changes of

vulnerability ratings was determined. Between both OS-VI, the range in vulnerability

scores was ±12 and only one instance of an LSOA vulnerability rating changing by more

than one gradation, that is, from a low to high was found. Instead, most changes were more

subtle, with the vulnerability rating changing by only one gradation, for example, from

moderate-high to high. The vulnerability rating of Bowthorpe (E01026794) in Norwich

increased from a low-moderate rating in the National OS-VI to a high rating in the Local

OS-VI and represents a change in vulnerability score of ±3, suggesting that the LSOA is

far more vulnerable within a local context than it is relative to the rest of England and

Wales. Analysis found that this was due to the LSOA having above- average

unemployment amongst working-age residents as well as above-average claimants of

working-age benefits, residents with low qualifications and households with female lone

parents in full-time employment within the Local OS-VI. This highlights the level of detail

and the local-level context that the OS-VI can reveal.

The findings from the OS-VI correspond very closely to oft-cited poverty and deprivation

maps produced by, for example, the information services group Experian (Experian, 2014).

In particular, those areas with a high OS-VI rating correlate well with those areas

highlighted within the Experian maps (Rogers, 2012) as having the greatest overall risk of

poverty, the largest instances of child poverty, and the largest proportion of households

whose income is less than 60% of the median for England – all seen as indicators of

vulnerability (DETR, 2000; Hills, 2012; Whelan & Bertrand, 2005).

The pattern of vulnerability displayed within the OS-VI, a general radial configuration with

vulnerability decreasing with distance from major urban areas, was found to differ from

the pattern of deprivation displayed in the IMD 2010, a LSOA level measure of deprivation

produced by the University of Oxford for the UK Government’s Department for

Communities and Local Government (DCLG, 2011). Contrary to the OS-VI, the IMD

ranking within major urban areas, such as the city of Norwich, and several of the coastal

LSOA within the flood zone are characterised by relatively low deprivation. Similarly,

many of the suburban areas with low OS-VI ratings record high deprivation ratings within

the IMD. Although the OS-VI and IMD are measuring different concepts – vulnerability

and deprivation, respectively – the working definitions of the two concepts and the

indicators being measured are deemed sufficiently similar to warrant comparison. The

differences in the display of vulnerability and deprivation are undoubtedly due to the

methodological differences between the two indices and, in particular, the IMD’s focus on

income and employment deprivation and its exclusion of hazards and accessibility

measures as well as the equal weighting of categories within the OS-VI compared to the

subjective weightings of domains within the IMD. Combined, the Income Deprivation

Domain and the Employment Deprivation Domain within the IMD are weighted to account

for 45% of the overall IMD score. An examination of the variables within these domains,

all of which relate to families claiming forms of tax credits or claimants of unemployment -

related benefits amongst working- age residents, explains why areas along the coast that

are characterised by retired populations whose age and economic status likely exclude them

from those key IMD indicators record a relatively low IMD ranking, yet score a high OS-

VI rating. In addition, indices like the IMD more often than not focus on predominantly

urban concerns, such as unemployment, benefits claimants and poor housing, to the

detriment of rural concerns such as social isolation and accessibility (see: Farrington &

Farrington, 2005; Oxford Consultants for Social Inclusion, 2012; Preston & Rajé, 2007).

Figure 9. (a) Local OS-VI Norwich City detail and (b) National OS-VI Norwich City detail.

We see the OS-VI as a complementary tool to the IMD. The OS-VI is more inclusive,

taking into account a broader range of economic as well as social indicators that address

vulnerability in both urban and rural areas, and provides greater local context than the IMD,

highlighting potential vulnerability hotspots and the driving forces or root causes of

vulnerability.

6. Concluding remarks

There have been few examinations of comparative indicators of social vulnerability that

incorporate measures of accessibility and flood risk. Fewer studies still have integrated a

broad spectrum of vulnerability indicators available at the national level but with a

resolution that allows for the representation of local-level vulnerability. Furthermore,

limited attention has been paid to the use of open-source data and technology, with

proprietary sources preferred. The approach presented here addresses these limitations.

The index produced draws considerable influence from Blaikie et al.’s (1994) Pressure and

Release Model by directing attention to the root causes of social vulnerability within

communities but also Cutter et al.’s (1996, 2008) hazards of place model by focussing

explicitly on place vulnerability within the OS-VI, particularly relative local vulnerability

within the Local OS-VI. The inclusion of both vulnerability indicators and a measure of

risk potential via the flood hazard zone within the OS-VI provides a comprehensive picture

of social vulnerability and allows for the examination of the interaction between socio-

economic and biophysical vulnerabilities.

The goal of this project was to utilise free and readily available secondary data to produce

a tool that could be used by local councils or NGOs to identify communities that may

require added assistance before, during or after a flood event. The methodological approach

presented provides a mechanism whereby quality data on core drivers of vulnerability can

be used to create a vulnerability index that provides information at a national level but at a

sufficiently fine resolution so as to identify pockets of vulnerable communities. The

methods used are scalable and adaptable and the project’s reliance on open-source data and

technology significantly reduces the associated costs and allows all parties involved to

easily coordinate and share information, potentially improving local knowledge and

reducing vulnerability (Trujillo, Ordones, & Hernandes, 2000).

The OS-VI is the first step in imagining a dynamic and customisable platform that can

provide added context to complex situations and the targeting of resource and service

allocation, be it the provision of programmes to address an identified vulnerability stressor

or the location of new facilities to improve accessibility. Future work will focus on further

refining the variables and indicators used as well as examining the underlying dimensions

of social vulnerability and risk and its changes over space and time.

Disclosure statement

No potential conflict of interest was reported by the authors. Funding

This work was supported by the Engineering and Physical Sciences Research Council (EPSRC)

under Grant EP/G037698/1 and the British Red Cross (BRC).

ORCID

K. Garbutt http://orcid.org/0000-0001-9543-2110

C. Ellul http://orcid.org/0000-0002-9791-0259

References

Aday, L. A. (2001). At risk in America: The health and health care needs of vulnerable populations

in the United States. San Francisco, CA: Jossey-Bass.

Adger, W. N. (2006). Vulnerability. Global Environmental Change, 16(3), 268–281.Alexander,

D. (1993). Natural disasters. New York, NY: Chapman & Hall.

Anand, S., & Sen, A. (1994). Human development index: Methodology and measurement (No.

HDOCPA-1994–02). New York, NY: Human Development Report Office (HDRO), United

Nations Development Programme (UNDP).

Atteslander, P., Cromm, J., Grabow, B., Klein, H., Maurer, A., & Siegert, G. (2008). Methoden

der empirischen Sozialforschung. Berlin: Walter de Gruyter.

Auf der Heide, E., & Scanlon, J. (2007). Health and medical preparedness and response. In W. L.

Waugh & K. Tierney (Eds.), Emergency management: Principles and practice for local

government (pp. 183– 206). Washington, DC: International City Managers Association.

Balica, S., & Wright, N. G. (2010). Reducing the complexity of the flood vulnerability index.

Environmental Hazards, 9(4), 321–339.

Barnett, J., Lambert, S., & Fry, I. (2008). The hazards of indicators: Insights from the

environmental vulnerability index. Annals of the Association of American Geographers, 98(1),

102–119.

Bennet, G. (1970). Bristol floods 1968: Controlled survey of effects on health of local community

disaster.

British Medical Journal, 3, 454–458.Bianchi, S. M., & Spain, D. (1996). Women, work, and

family in America. Population Bulletin, 51(3), 1–48.

Bird, K., & Prowse, M. (2008). Vulnerability, poverty and coping in Zimbabwe (WIDER Research

Paper 2008/41). Helsinki: UNU-WIDER.

Birkmann, J. (2005). Report on the 1st meeting of the expert working group on ‘measuring

vulnerability’ (Working Paper No. 1). Bonn: United Nations University, UNU EHS.

Blaikie, P., Cannon, T., Davies, I., & Wisner, B. (1994). At risk: Natural hazards, people’s

vulnerability and disasters. New York, NY: Routledge.

Bohringer, C., & Jochem, P. E. P. (2007). Measuring the immeasurable: A survey of sustainability

indices. Ecological Economics, 63(1), 1–8.

Brooks, N., Adger, N., & Kelly, M. (2005). The determinants of vulnerability and adaptive

capacity at the national level and the implications for adaptation. Global Environmental Change,

15(2), 151–163.

Bruneau, M., Chang, R. T., Eguchi, G. C., Lee, T. D., O’Rourke, A. M., Reinhorn, M., ... von

Winterfeldt, D. (2003). A framework to quantitatively assess and enhance the seismic resilience

of communities. Earthquake Spectra, 19(4), 733–752.

Burd-Sharps, S., Lewis, K., & Martins, E. B., (Eds.). (2008). The measure of America: American

human development report, 2008–2009. New York, NY: Columbia University Press.

Cabinet Office. (2013). National risk register of civil emergencies 2013 edition. London: Author.

Campbell, N. C., Elliott, A. M., Sharp, L., Ritchie, L. D., Cassidy, J., & Little, J. (2000). Rural

factors and survival from cancer: Analysis of Scottish cancer registrations. British Journal of

Cancer, 82, 1863–1866.

Carroll, A. (2012, May 14). Map shows Norfolk flooding risk zones. Eastern Daily Press.

Chen, A., Keith, V. M., Airriess, C., Li, W., & Leong, K. J. (2007). Economic vulnerability,

discrimination, and Hurricane Katrina: Health among black Katrina survivors in eastern New

Orleans. Journal of the American Psychiatric Nurses Association, 13(5), 257–266.

Coffee, N., Turner, D., Clark, R. A., Eckert, K., Coombe, D., Hugo, G., & Tonkin, A. A. (2012).

Measuring national accessibility to cardiac services using geographic information systems.

Applied Geography, 34, 445–455.

Comber, A., Brunsdon, C., & Green, E. (2008). Using a GIS-based network analysis to determine

urban greenspace accessibility for different ethnic and religious groups. Landscape and Urban

Planning, 86 (1), 103–114.

Cox, J. R., Rosenzweig, C., Solecki, W., Goldberg, R., & Kinney, P. (2007). Social vulnerability

to climate change: A neighborhood analysis of the northeast US Megaregion. Technical paper

prepared for the Northeast Climate Impacts Assessment, Cambridge, MA.

Cutter, S. (1996). Vulnerability to environmental hazards. Progress in Human Geography, 20(4),

529–539. Cutter, S. (2006, June 11). The geography of social vulnerability: Race, class, and

catastrophe. Retrieved from http://understandingkatrina.ssrc.org/Cutter/

Cutter, S., Barnes, L., & Berry, M. (2008). Community and regional resilience: Perspectives form

hazards, disasters and emergency management community and regional resilience initia t ive

(CARRI) (Research Report 1). Oak Ridge: Community and Regional Resilience Institute.

Retrieved from http://www. resilientus.org/library/FINALCUTTER9–25–08_1223482309.pdf

Cutter, S., Boruff, B., & Shirley, W. (2003). Social vulnerability to environmental hazards. Social

Science Quarterly, 84(2), 242–261.

Cutter, S., Burton, C. G., & Emrich, C. T. (2010). Disaster resilience indicators for benchmarking

baseline conditions. Journal of Homeland Security and Disaster Resilience, 7(1), 1–22.

Cutter, S. L., & Emrich, C. T. (2006). Moral hazard, social catastrophe: The changing face of

vulnerability along the hurricane coasts. The Annals of the American Academy of Political and

Social Science, 604(1), 102–112.

Cutter, S., Emrich, C. T., Webb, J. J., & Morath, D. (2009). Social vulnerability to climate

variability hazards: A review of the literature (Final Report to Oxfam America, 5, 1–44).

Columbia: University of South Carolina.

Cutter, S. L., Mitchell, J. T., & Scott, M. S. (2000). Revealing the vulnerability of people and

places: A case study of Georgetown County, South Carolina. Annals of the Association of

American Geographers, 90 (4), 713–737.

Damm, M. (2010). Mapping social-ecological vulnerability to flooding. A sub-national approach

for Germany (Graduate Research Series vol. 3). Bonn: UNU-EHS.

Delafontaine, M., Neutens, T., & Weghe, N. (2012). A GIS toolkit for measuring and mapping

space – Time accessibility from a place-based perspective. International Journal of Geographica l

Information Science, 26(6), 1131–1154.

Department for Communities and Local Government. (2011). English indices of deprivation 2010.

London: Author.Department for Transport. (2004). Technical guidance on accessibility planning

in local transport plans. London: Author.

Department for Transport. (2012). Accessibility statistics 2011. London: Author.DETR. (2000).

Measuring multiple deprivation at the small area level: The indices of deprivation 2000. London:

Author.

Dibben, C. Atherton, I., Cox, M., Watson, V., Ryan, M., & Sutton, M. (2004). Investigating the

impact of changing the weights that underpin the index of multiple deprivation 2004. London:

Department for Communities and Local Government.

Dwyer, A., Zoppou, C., Nielsen, O., Day, S., & Roberts, S. (2004). Quantifying social

vulnerability: A methodology for identifying those at risk to natural hazards (Geoscience Australia

Record 14, 1–92). Canberra: Geoscience Australia.

Enarson, E. (2007). Identifying and addressing social vulnerabilities. In K. Tierney & K. Waugh

(Eds.), Emergency management: Principles and practice for local government (pp. 257–278).

Washington, DC: International City and County Management Association.

Enarson, E., & Morrow, B. (1998). The gendered terrain of disaster. New York: Praeget.Engle,

N. L. (2011). Adaptive capacity and its assessment. Global Environmental Change, 21(2), 647–

656.

Erskine, S. (2010). Socioeconomic deprivation, urban-rural location and alcohol-related mortality

in England and Wales. BMC Public Health, 10(1), 1–99.

Esty, D., Levy, M., Srebotnjak, T., & de Sherbinin, A. (2005). Environmental sustainability index:

Benchmarking national environmental stewardship. New Haven, CT: Yale Center for

Environmental Policy and Law.Experian. (2014). MOSAIC classification. Retrieved from

http://www.experian.co.uk/marketing-services/products/mosaic-uk.html

Farrington, J., & Farrington, C. (2005). Rural accessibility, social inclusion and social justice: Towards conceptualisation. Journal of Transport Geography, 13(1), 1–12.

Fekete, A. (2011). Spatial disaster vulnerability and risk assessments: Challenges in their quality

and acceptance. Natural Hazards, 61(3), 1161–1178.

Floyd, P., & Tunstall, S. (Risk and Policy Analysis/Flood Hazard Research Centre). (2004). The

appraisal of human-related intangible impacts of flooding (Report to Defra/Environment Agency,

R&D Project FD2005). London: Defra.

Garbutt, K. (2013). Using GIS and an indicator-based approach to identify and map social

vulnerability: A case study of Norfolk, England. A thesis presented for the degree of Master of

Research at University College London.

Gartner, A. (2011). Rural/urban mortality differences in England and Wales and the effect of

deprivation adjustment. Social Science & Medicine, 72(10), 1685–1694.

Goss, J. (1995). We know who you are and we know where you live: The instrumental rationality of geo- demographic systems. Economic Geography, 71(2), 171–198.

Greater London Authority. (2013, November). Better environment, better health: A GLA guide

for London’s Boroughs London Borough of Waltham Forest. London: Author.

Gunderson, L. (2009). Comparing ecological and human community resilience (CARRI Research

Report 5). Oak Ridge: Community and Regional Resilience Institute.

Hajat, S., Ebi, K. L., Kovats, R. S., Menne, B., Edwards, S., & Haines, A. (2005). The human

health consequences of flooding in Europe: A review. In Extreme weather events and public health

responses (pp. 185–196). Berlin Heidelberg: Springer.

Haki, Z., & Akyurek, Z. (2004, April 29–May 1). Assessment of social vulnerability using

geographic infor- mation systems: Pendik, Istanbul case study. Paper presented at the 7th AGILE

conference on Geographic Information Science, Heraklion, Crete.

Halden, D., Jones, P., & Wixey, S. (2005). Measuring accessibility as experienced by different

socially dis- advantaged groups: Accessibility analysis literature review. London: Transit Studies

Group, University of Westminster.

Higgs, G. (2004). A literature review of the use of GIS-based measures of access to health care

services. Health Services & Outcomes Research Methodology, 5(2), 119–139.

Hills, J. (2012). Getting the measure of fuel poverty final report of the fuel poverty review. CASE report 72 Commissioned by the UK Government’s Department of Energy and Climate Change

(DECC).

Jones, P. G., & Thornton, P. K. (2003). The potential impacts of climate change on maize production in Africa and Latin America in 2055. Global Environmental Change, 13, 51–59.

Kasperson, J. X., & Kasperson, R. E. (2001). International workshop on vulnerability and global

environ- mental change (SEI Risk and Vulnerability Programme Report 2001–01). Stockholm: Stockholm Environment Institute.

Kazmierczak, A., & Cavan, G. (2011). Surface water flooding risk to urban communities: Analys is

of vulnerability, hazard and exposure. Landscape and Urban Planning, 103(2), 185–197.

Kværner, J., Swensen, G., & Erikstad, L. (2006). Assessing environmental vulnerability in EIA –

the content and context of the vulnerability concept in an alternative approach to standard EIA

procedure. Environmental Impact Assessment Review, 26(5), 511–527.

Lankao, P. A., & Qin, H. (2011). Conceptualizing urban vulnerability to global climate and

environmental change. Current Opinion in Environmental Sustainability, 3(3), 142–149.

Liu, S., & Zhu, X. (2004). An integrated GIS approach to accessibility analysis. Transactions in

GIS, 8, 45–62. Liverman, D. (1989). Vulnerability to global environmental change. In R. E.

Kasperson, K. Dow, D. Golding, & J. X. Kasperson (Eds.), Understanding global environmenta l

change: The contributions of risk analysis and management (pp. 27–44). Worcester, MA: Clark

University, The Earth Transformed Program 1.

Long, A. (2007). Poverty is the new prostitution: Race, poverty, and public housing in post-Katrina

New Orleans. Journal of American History, 94, 795–803.

Lucas, K., Tyler, S., & Christodoulou, G. (2009). Assessing the ‘value’ of new transport initiat ives

in deprived neighbourhoods in the UK. Transport Policy, 16(3), 115–122.

Marmot, M. (2005). Social determinants of health inequalities. The Lancet, 365(9464), 1099–

1104.

Mavoa, S., Witten, K., McCreanor, T., & O’Sullivan, D. (2012). GIS based destination

accessibility via public transit and walking in Auckland, New Zealand. Journal of Transport

Geography, 20(1), 15–22.

Meyer, V., Haase, D., & Scheuer, S. (2007). GIS-based multicriteria analysis as decision support

in flood risk management (Integrated project FLOODsite, milestone report). Helmho lz

Unweltforschungszentrum (UFZ). Department of Economics, Leipzig.

Miller, H. J. (2003). Transportation and communication lifeline disruption. In S. L. Cutter, D. B.

Richardson, & T. Wilbanks (Eds.), The geographic dimensions of terrorism (pp. 145–152).

Evanston, IL: Routledge.

Moldan, B., & Dahl, A. L. (2007). Challenges to sustainability indicators. In T. Hak, B. Moldan,

& A. L. Dahl (Eds.), Sustainability indicators. A scientific assessment (pp. 1–27). Washington,

DC: Island Press.

Morath, D. P. (2010). Social vulnerability and public health: Developing a metric for medical

emergency management in Florida. A thesis presented for the degree of Master of Science at the

University of South Carolina.

Morrow, B. (1999). Identifying and mapping community vulnerability. Disasters, 23(1), 1–

18.Morrow, B. (2008). Community resilience: A social justice perspective (The community and

regional resilience initiative research report 4). Oak Ridge, TN: Community and Regional

Resilience Initiative.

Morse, S. (2004). Indices and indicators in development: An un-healthy obsession with numbers.

Sterling, VA: Earthscan.

Mustafa, D., Ahmed, S., Saroch, E., & Bell, H. (2011). Pinning down vulnerability: From

narratives to numbers. Disasters, 35(1), 62–86.

Musterd, S., & Ostendorf, W. (2013). Urban segregation and the welfare state: Inequality and

exclusion in western cities. New York, NY: Routledge.

Müller, A., Reiter, J., & Weiland, U. (2011). Assessment of urban vulnerability towards floods

using an indicator-based approach – A case study for Santiago de Chile. Natural Hazards and Earth

System Science, 11(8), 2107–2123.

Nelson, C., Lurie, N., Wasserman, J., Zakowski, S., & Leuschner, K. J. (2007). Conceptualizing

and defining public health emergency preparedness (RAND Working Paper – May 2008).

Emmitsburg, MD: National Emergency Training Center.

Nettleton, M., Pass, D. J., Walters, G. W., & White, R. C. (2007). Public transport accessibility

map of access to general practitioners surgeries in Longbridge, Birmingham, UK. Journal of Maps,

2, 64–75.

Neutens, T., Delafontaine, M., Scott, D. M., & De Maeyer, P. (2012). A GIS-based method to

identify spatio-temporal gaps in public service delivery. Applied Geography, 32(2), 253–264.

Nicholl, J., West, J., Goodacre, S., & Turner, J. (2007). The relationship between distance to

hospital and patient mortality in emergencies: An observational study. Emergency Medicine

Journal, 24(9), 665–668.

Niemeijer, M. (2002). Developing indicators for environmental policy: Data-driven and theory-

driven approaches examined by example. Environmental Science & Policy, 5, 91–103.

Ngamini Ngui, A., & Vanasse, A. (2012). Assessing spatial accessibility to mental health facilit ies

in an urban environment. Spatial and Spatio-Temporal Epidemiology, 3(3), 195–203.

Norris, F. H., Stevens, S. P., Pfefferbaum, B., Wyche, K. F., & Pfefferbaum, R. L. (2008).

Community resilience as a metaphor, theory, set of capacities, and strategy for disaster readiness.

American Journal of Community Psychology, 41, 127–150.

Office for National Statistics. (2013). Census key statistics. London: Author.

O’Sullivan, C., Mulgan, G., Ali, R., & Norman, W. (2009). Sinking and swimming: Understanding

Britain’s unmet needs. London: Young Foundation.

Oxfam America. (2009). Exposed: Social vulnerability and climate change in the US Southeast.

Boston, MA: Oxfam.

Oxford Consultants for Social Inclusion. (2012). Getting the measure of rural deprivation in Wales:

A report commissioned by the Local Government Data Unit, Wales Office for National Statistics,

2013. Cardiff: Local Government Data Unit.

Páez, A., Scott, D. M., & Morency, C. (2012). Measuring accessibility: Positive and normative

implemen- tations of various accessibility indicators. Journal of Transport Geography, 25, 141–

153.

Patt, A. G., Tadross, M., Nussbaumer, P., Asante, K., & Metzger, M. (2010). Estimating least-

developed countries’ vulnerability to climate-related extreme events over the next 50 years. PNAS,

107(4), 1333–1337.

Peguero, A. A. (2008). The victimization and vulnerabilities of immigrant children: The

importance of English proficiency. In R. Perrucci, K. Ferraro, J. Miller, & G. W. Muschert (Eds.),

Agenda for social justice: Solutions 2008. Knoxville, TN: Society for the Study of Social

Problems.

Polsky, C., Neff, R., & Yarnal, B. (2007). Building comparable global change vulnerabil ity

assessments: The vulnerability scoping diagram. Global Environmental Change, 17, 472–485.

Prescott-Allen, R. (2001). The wellbeing of nations. New York, NY: Island Press.Preston, B. L.

(2012). Climate change vulnerability assessment: From conceptual frameworks to practical

heuristics (Climate Adaptation Flagship Working Paper). Dickson, CA: CSIRO.

Preston, J., & Rajé, F. (2007). Accessibility, mobility and transport-related social exclusion.

Journal of Transport Geography, 15(3), 151–160.

Reacher, M., McKenzie, K., Lane, C., Nichols, T., Iversen, A., Hepple, P., ... Simpson, J. (2004).

Health impacts of flooding in Lewes: A comparison of reported gastrointestinal and other illness

and mental health in flooded and non-flooded households. Communicable Disease and Public

Health, 7(1), 1–8.

Ribeiro, A., Antunes, A. P., & Páez, A. (2010). Road accessibility and cohesion in lagging regions:

Empirical evidence from Portugal based on spatial econometric models. Journal of Transport, 10,

125–132.

Rogers, S. (2012, March 6). Poverty in England: How Experian ranks where you live. Guardian.

Retrieved from http://www.theguardian.com/news/datablog/2012/mar/06/poverty-england-

experian-rankings

Rygel, L., O’Sullivan, D., & Yarnal, B. (2006). A method for constructing a social vulnerability

index: An application to hurricane storm surges in a developed country. Mitigation and Adaptation

Strategies for Global Change, 11(3), 741–764.

Schneiderbauer, S. (2007). Risk and vulnerability to natural disasters–from broad view to focused

perspective: Theoretical background and applied methods for the identification of the most

endangered populations in two case studies at different scales. A thesis presented for the degree of

Doctor of Philosophy at Freie Universität, Berlin.

Scott, D. M., & Horner, M. W. (2008). Examining the role of urban form in shaping people’s

accessibility to opportunities: An exploratory spatial data analysis. Journal of Transport and Land

Use, 1(2), 89–119.

Stânga, I. C., & Grozavu, A. (2012). Quantifying human vulnerability in rural areas: Case study

of Tutova Hills (Eastern Romania). Natural Hazards and Earth System Science, 12(6), 1987–2001.

Tapsell, S. M., & Tunstall, S. M. (2006). The mental health aspects of floods: Evidence from

England and Wales. In R. Few & F. Matties (Eds.), Flood hazards and health: Responding to

present and future risks (pp. 89–110). London: Earthscan.

Tate, E. (2013). Uncertainty analysis for a social vulnerability index. Annals of the Association of

American Geographers, 103(3), 526–543.

Taubenböck, H., Post, J., Roth, A., Zosseder, K., Strunz, G., & Dech, S. (2008). A conceptual

vulnerability and risk framework as outline to identify capabilities of remote sensing. Natural

Hazards and Earth System Science, 8(3), 409–420.

Tierney, K. J., Lindell, M. K., & Perry, R. W. (2001). Facing the unexpected: Disaster

preparedness and response in the United States. Washington, DC: Joseph Henry Press.

Trujillo, M., Ordones, A., & Hernandes, C. (2000). Risk mapping and local capacities: Lessons

from Mexico and Central America. Oxford: Oxfam.

Trujillo-Pagan, N. (2007). Katrina’s Latinos: Vulnerability and disasters in relief and recovery. In

K. A. Bates & R. S. Swan (Eds.), Through the eye of Katrina: Social justice in the United States

(pp. 147–168).

Durham, NC: Carolina Academic Press.Tunstall, S. M., Tapsell, S. M., Green, C., Floyd, P., &

George, C. (2006). The health effects of flooding: Social research results from England and Wales.

Journal of Water and Health, 4(3), 365–380.

Turvey, R. (2007). Vulnerability assessment of developing countries: The case of small island

developing states. Development Policy Review, 25(2), 243–264.

Van Zandt, S., Peacock, W. G., Henry, D. W., Grover, H., Highfield, W. E., & Brody, S. D. (2012).

Mapping social vulnerability to enhance housing and neighborhood resilience. Housing Policy

Debate, 22(1), 29–55.

Vincent, K. (2004). Creating an index of social vulnerability to climate change for Africa (Tyndall

Center for Climate Change Research Working Paper 56, p. 41). Norwich: Tyndall Centre for

Climate Change Research.

Wamsler, C. (2014). Cities. Disaster risk and adaptation. New York, NY: Routledge.

Watts, M. J., & Bohle, H. G. (1993). The space of vulnerability: The causal structure of hunger

and famine. Progress in Human Geography, 17(1), 43–67.

Webber, D. J., & Rossouw, S. (2010). Sub-national vulnerability measures: A spatial perspective,

Auckland (Economics Discussion Paper Series, 1004). Bristol: Department of Accounting,

Economics and Finance, Bristol Business School, University of the West of England.

Whelan, C. T., & Bertrand, M. (2005). Vulnerability and multiple deprivation perspectives on

economic exclusion in Europe: A latent class analysis. European Societies, 7(3), 423–450.

Wisner, B., Blaikie, P., Cannon, T., & Davis, I. (2004). At risk: Natural hazards, people’s

vulnerability and disasters. Oxford: Routledge.World Health Organization. (2002). Floods:

Climate change and adaptation strategies for human health (Meeting report). London, 30 June–2

July 2002. Copenhagen: WHO Regional Office for Europe. Retrieved from

http://www.who.dk/document/E77096.pdf

Yoon, D. K. (2012). Assessment of social vulnerability to natural disasters: A comparative study.

Natural Hazards, 63(2), 823–843.